Circle Intersection

Don't know how to use a machine? Looking for efficient setups? Stuck in a mission?
Post Reply
Mr. Tact
Filter Inserter
Filter Inserter
Posts: 460
Joined: Sat Mar 26, 2016 3:37 pm
Contact:

Circle Intersection

Post by Mr. Tact »

Is there going to be any noticeable difference in the throughput of these intersections?

circle 1.jpg
circle 1.jpg (92 KiB) Viewed 8230 times
circle 2.jpg
circle 2.jpg (94.87 KiB) Viewed 8230 times
Professional Curmudgeon since 1988.

Loewchen
Global Moderator
Global Moderator
Posts: 8283
Joined: Wed Jan 07, 2015 5:53 pm
Contact:

Re: Circle Intersection

Post by Loewchen »

There is no difference at all. There is no point in splitting this block because no possible train path could occupy only one of the now two blocks.

Amarula
Filter Inserter
Filter Inserter
Posts: 509
Joined: Fri Apr 27, 2018 1:29 pm
Contact:

Re: Circle Intersection

Post by Amarula »

I have tried both ways, in fact I was using version 2 as my standard until I took a closer look just before Christmas.
Version 1 divides the circle into quarters; version 2 divides it into eighths, where one signal guards the entrance to the quarter, and the next guards the exit from the quarter.
Version 2 lets trains enter the circle slightly sooner, as they don't have to wait for the previous train to exit the circle, just exit the entrance to the quarter. So yes higher throughput - noticeable, not so sure.

Reasons why I stopped using version 2:
1. More signals means more times the trains potentially stop and recalculate the path - I have watched a train go completely around the circle before taking an exit, which kills any throughput gain really fast.
2. Why guard an exit from a segment? There is no possibility of a collision, because there is only ever one train in a segment. Guarding the exit adds complexity without adding any protection.
3. Splitting the circle into eighths means that a train, even a one locomotive - one cargo train, doesn't fit into a segment, it is always blocking two segments (one entrance to a quarter and one exit). Again extra complexity without corresponding benefit to justify it (as far as I can tell).
4. Twice, in about 250 hours of play, in a high traffic area (15+ trains using one intersection), I managed to deadlock trains in the circle - 4 trains using the circle, and none of them able to exit. I still don't understand why, changing one train to manual and driving it out of the circle cleared the problem - I didn't have to move any other trains out of the way.

Great question, thanks for asking it!
My own personal Factorio super-power - running out of power.

Mr. Tact
Filter Inserter
Filter Inserter
Posts: 460
Joined: Sat Mar 26, 2016 3:37 pm
Contact:

Re: Circle Intersection

Post by Mr. Tact »

Loewchen wrote:
Thu Jan 10, 2019 3:07 pm
There is no difference at all. There is no point in splitting this block because no possible train path could occupy only one of the now two blocks.
That's what I was thinking, but I couldn't manage to convince myself I was right. Thanks.
Professional Curmudgeon since 1988.

Mr. Tact
Filter Inserter
Filter Inserter
Posts: 460
Joined: Sat Mar 26, 2016 3:37 pm
Contact:

Re: Circle Intersection

Post by Mr. Tact »

Amarula wrote:
Thu Jan 10, 2019 3:33 pm
4. Twice, in about 250 hours of play, in a high traffic area (15+ trains using one intersection), I managed to deadlock trains in the circle - 4 trains using the circle, and none of them able to exit. I still don't understand why, changing one train to manual and driving it out of the circle cleared the problem - I didn't have to move any other trains out of the way.
I've never had a deadlock in a circle intersection I couldn't explain for some other reason (like a train running out of fuel or a mistake in the signaling), but I've occasionally read circle intersections are supposedly "not efficient" and I was watching a section of busy track and suddenly wondered if this would change anything. And couldn't convince myself either way.

A related question -- when used for only three directions of traffic does a circle intersection work better or worse or the same as something like this:

3 way.jpg
3 way.jpg (497.25 KiB) Viewed 8206 times
Once again I am unable to convince myself either is noticeably different in throughput capacity.
Professional Curmudgeon since 1988.

Serenity
Smart Inserter
Smart Inserter
Posts: 1000
Joined: Fri Apr 15, 2016 6:16 am
Contact:

Re: Circle Intersection

Post by Serenity »

That's way too many chain signals on that T-junction

Mr. Tact
Filter Inserter
Filter Inserter
Posts: 460
Joined: Sat Mar 26, 2016 3:37 pm
Contact:

Re: Circle Intersection

Post by Mr. Tact »

Hmm, it is a blueprint I've had a long time. But looking at it with a signal in my hand:

3 way-2.jpg
3 way-2.jpg (386.09 KiB) Viewed 8197 times

All the cross or merge points are separate blocks, isn't that what you want?
Professional Curmudgeon since 1988.

Loewchen
Global Moderator
Global Moderator
Posts: 8283
Joined: Wed Jan 07, 2015 5:53 pm
Contact:

Re: Circle Intersection

Post by Loewchen »

Mr. Tact wrote:
Thu Jan 10, 2019 5:30 pm
All the cross or merge points are separate blocks, isn't that what you want?
You only need chain signals when a train passing that signal and stopping on the next one would block A DIFFERENT PATH than the path passing this signal, so no chain signals for merging: "T" for normal train signal.
T-jun__sig_fac.png
T-jun__sig_fac.png (1.63 MiB) Viewed 8193 times

Mr. Tact
Filter Inserter
Filter Inserter
Posts: 460
Joined: Sat Mar 26, 2016 3:37 pm
Contact:

Re: Circle Intersection

Post by Mr. Tact »

Hmm, okay. Guess it is time to do some updates...

And as to whether this T or the circle has a higher throughput? No difference?
Professional Curmudgeon since 1988.

User avatar
DaveMcW
Smart Inserter
Smart Inserter
Posts: 3699
Joined: Tue May 13, 2014 11:06 am
Contact:

Re: Circle Intersection

Post by DaveMcW »

The circle and all-chain-signal T should have the same throughput.

The T with optimal signals will be slightly better.

Mr. Tact
Filter Inserter
Filter Inserter
Posts: 460
Joined: Sat Mar 26, 2016 3:37 pm
Contact:

Re: Circle Intersection

Post by Mr. Tact »

Okay, I just want to make sure I understand this. If you make the signal changes indicated by Loewchen, that makes the T slightly faster why? Fewer recalculations due to fewer chain signals?
Professional Curmudgeon since 1988.

Loewchen
Global Moderator
Global Moderator
Posts: 8283
Joined: Wed Jan 07, 2015 5:53 pm
Contact:

Re: Circle Intersection

Post by Loewchen »

Mr. Tact wrote:
Thu Jan 10, 2019 7:21 pm
Okay, I just want to make sure I understand this. If you make the signal changes indicated by Loewchen, that makes the T slightly faster why? Fewer recalculations due to fewer chain signals?
It's about throughput of the junction not game performance, chain signals force a larger gap between trains, if you use them in places where they are not needed the only effect is that trains wait unnecessarily.

Mr. Tact
Filter Inserter
Filter Inserter
Posts: 460
Joined: Sat Mar 26, 2016 3:37 pm
Contact:

Re: Circle Intersection

Post by Mr. Tact »

Ah, okay. Cool. Thanks.
Professional Curmudgeon since 1988.

Mr. Tact
Filter Inserter
Filter Inserter
Posts: 460
Joined: Sat Mar 26, 2016 3:37 pm
Contact:

Re: Circle Intersection

Post by Mr. Tact »

Actually, thinking about it some more, based on what you said... shouldn't the signals next to the green Ts I've inserted also be changed to regular instead of chain signals?

"no chain signals for merging" ... So, basically -- the only time you want a chain signal is when you are crossing another rail. Not prior to a fork and not prior to a merge. Right? Wow, if that's right, I just changed my basic understanding of rail signals after playing this game for over 1500 hours.

modified T.jpg
modified T.jpg (297.26 KiB) Viewed 8145 times
Professional Curmudgeon since 1988.

Jap2.0
Smart Inserter
Smart Inserter
Posts: 2339
Joined: Tue Jun 20, 2017 12:02 am
Contact:

Re: Circle Intersection

Post by Jap2.0 »

Amarula wrote:
Thu Jan 10, 2019 3:33 pm
Version 2 lets trains enter the circle slightly sooner, as they don't have to wait for the previous train to exit the circle, just exit the entrance to the quarter. So yes higher throughput - noticeable, not so sure.
No, it actually won't help throughput, because while the train can enter when it clears that block, it won't until the block it needs to exit from is clear - and the endings of those blocks are at the same place in both circles.
There are 10 types of people: those who get this joke and those who don't.

Loewchen
Global Moderator
Global Moderator
Posts: 8283
Joined: Wed Jan 07, 2015 5:53 pm
Contact:

Re: Circle Intersection

Post by Loewchen »

Mr. Tact wrote:
Thu Jan 10, 2019 9:24 pm
Actually, thinking about it some more, based on what you said... shouldn't the signals next to the green Ts I've inserted also be changed to regular instead of chain signals?
Exactly right, but only if you have very short trains with one or no wagons, otherwise trains would be able to enter the junction and stop at the normal signals after the green Ts which are so close that the rear would block the yellow block in the middle.
Mr. Tact wrote:
Thu Jan 10, 2019 9:24 pm
"no chain signals for merging" ... So, basically -- the only time you want a chain signal is when you are crossing another rail. Not prior to a fork and not prior to a merge. Right? Wow, if that's right, I just changed my basic understanding of rail signals after playing this game for over 1500 hours.
Yes. Chain signals are historically misused, when they were first introduced deadlocks were the big sickness of the time and people knew chain signals could prevent them, just not how exactly, so they replaced all junction signals with them and since this worked to prevent deadlocks they kept doing it. Somewhat like antibiotics use in the 60s.

Post Reply

Return to “Gameplay Help”